Why can't AMD keep Keller around?

etherealfocus

Senior member
Jun 2, 2009
488
13
81
http://fudzilla.com/news/43809-keller-is-building-self-driving-tesla

Seriously, this pattern keeps happening:

1. AMD completely uncompetitive, barely afloat, bleeding money

2. Keller shows up

3. AMD launches new CPU, back in the game

4. Keller leaves

5. AMD milks his work until it's woefully outdated, then releases a new chip that's barely better than Keller's work despite years of dev time

6. Back to Step 1

Are they just not paying him enough? Does Keller only want to build that revolutionary chip every decade or so and then take off again for some reason?
 

Yakk

Golden Member
May 28, 2016
1,574
275
81
IMHO Jim Keller is AAAAA talent who can afford to do what he wants. It's not just AMD who can't hold on to him, he doesn't stay anywhere very long... DEC, AMD, Apple (their A processors) , Samaung(rumor) , and now Telsa.
 

scannall

Golden Member
Jan 1, 2012
1,960
1,678
136
http://fudzilla.com/news/43809-keller-is-building-self-driving-tesla

Seriously, this pattern keeps happening:

1. AMD completely uncompetitive, barely afloat, bleeding money

2. Keller shows up

3. AMD launches new CPU, back in the game

4. Keller leaves

5. AMD milks his work until it's woefully outdated, then releases a new chip that's barely better than Keller's work despite years of dev time

6. Back to Step 1

Are they just not paying him enough? Does Keller only want to build that revolutionary chip every decade or so and then take off again for some reason?
He seems to have wanderlust. And that's fine. Goes from project to project that interests him.
 
  • Like
Reactions: NTMBK

ddogg

Golden Member
May 4, 2005
1,864
361
136
I think he enjoys working on something new and revolutionary, rather than being in one company working on the same thing. Just my guess.
 

formulav8

Diamond Member
Sep 18, 2000
7,004
522
126
While Keller is A+, you should give a lot of credit to Mark Papermaster. Keller more so helped to put together and organize the team, while Papermaster headed the project.
 

moinmoin

Diamond Member
Jun 1, 2017
5,236
8,443
136
He stayed around longer this time than last time actually. (Left 1999, K8 released 2003, left 2015, Zen released 2017.) He left respectively when his job was done.
 

Dribble

Platinum Member
Aug 9, 2005
2,076
611
136
This Keller worship is silly. He didn't stay around for Zen, it was 2 years after he left it was released, that's not "staying around". I'm sure he's a bright guy, I'm sure he did some good stuff, but he was just one of many bright guys. If you want to know why Ryzen is good, it's probably because a number of bright guys you've never heard off worked hard and made it good, not some famous name working at a management level.
 

wildhorse2k

Member
May 12, 2017
180
83
71
I think that skilled and happy workforce is much more important than one top level manager. AMD surely must have enough competent people as they came up with Ryzen. These manages come, change things, cause chaos and before it settles down there is new manager who does the same again.
 

Snarf Snarf

Senior member
Feb 19, 2015
399
327
136
I don't remember the original source for this information, but there was a great article that described exactly what Keller's influence was on the CPU design team. Essentially they said he introduced a lot of modern testing and tooling techniques that AMD previously wasn't using. A lot of it was adding a ton of sensors into the CPU and introducing automated tools to make testing far more efficient than it was previously being done. This in turn allowed the engineers to spend more time on other aspects of the design process, and helps speed up the process of respins and stepping changes.

There's a good reason Keller is in high demand and can essentially go to any project he wants. That is because he has vision and the ability to turn that vision into actual products with any talent that he's given. Is he the god of CPU design? Hardly, but he is a very good leader of teams, and whatever he puts his hands on generally becomes a very good product.
 

turtile

Senior member
Aug 19, 2014
632
313
136
I don't remember the original source for this information, but there was a great article that described exactly what Keller's influence was on the CPU design team. Essentially they said he introduced a lot of modern testing and tooling techniques that AMD previously wasn't using. A lot of it was adding a ton of sensors into the CPU and introducing automated tools to make testing far more efficient than it was previously being done. This in turn allowed the engineers to spend more time on other aspects of the design process, and helps speed up the process of respins and stepping changes.

There's a good reason Keller is in high demand and can essentially go to any project he wants. That is because he has vision and the ability to turn that vision into actual products with any talent that he's given. Is he the god of CPU design? Hardly, but he is a very good leader of teams, and whatever he puts his hands on generally becomes a very good product.

Since he is always moving to new companies, he's always learning the good and the bad - building from the experience. Most who stay at the same company don't even realize how different another company works and can tend to get stuck doing the same thing the same way stubbornly.
 

etherealfocus

Senior member
Jun 2, 2009
488
13
81
I certainly don't want to diminish the work of the rank and file engineers, but clearly there was protracted bad mojo in the office for a very long time after K8. I dunno if they cut their team or mismanaged them or hired the wrong HR firm to replace lost talent or had morale issues or some other talented senior person took off or what. I follow this stuff pretty casually; just noticed that Keller seems to have been AMD's guardian angel several times here, and AMD would certainly love access to the markets Keller has been hitting for other companies - mobile, auto, etc. And they're certainly in dire need of consistent, strong evolutions rather than an Intel challenger once or twice a decade. Ryzen is awesome but if Ryzen 2 doesn't shine it's just postponing the inevitable.
 
  • Like
Reactions: Drazick

bononos

Diamond Member
Aug 21, 2011
3,928
186
106
This Keller worship is silly. He didn't stay around for Zen, it was 2 years after he left it was released, that's not "staying around". I'm sure he's a bright guy, I'm sure he did some good stuff, but he was just one of many bright guys. If you want to know why Ryzen is good, it's probably because a number of bright guys you've never heard off worked hard and made it good, not some famous name working at a management level.

Keller was the lead architect and I think he alone drew up the basic Ryzen design. He isn't some management exec. So he designed the CCX's and Infinity Fabric which are going to be re-used for the next gen products for cpu and gpu divisions.
 

Topweasel

Diamond Member
Oct 19, 2000
5,437
1,659
136
Keller was the lead architect and I think he alone drew up the basic Ryzen design. He isn't some management exec. So he designed the CCX's and Infinity Fabric which are going to be re-used for the next gen products for cpu and gpu divisions.
Nope Keller was brought in to help or head the K12 (ARM server CPU) he worked with AMD on picking out a team for the Zen design. He imparted wisdom and techniques and once his contract was up and K12 practically dead, he left.

Not only that but IF has as much influence if not more from the Radeon side. Part of the reason the it's bandwidth is where it is at on Ryzen is because it was designed to scale down from the 500Gb/s needed on the GPU down to the Ryzen Speeds.
 
  • Like
Reactions: Drazick